Unverified50% confidenceFactExact time
OmniCompass-VL contains 820 videos and 2460 instances, covering multiple domains and three duration types: short, medium, and long.
1
Sources
50%
Confidence
Long-term
Relevance
10/14/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedOmniCompass's core design philosophy is to require omni-modal video understanding to cover three dimensions simultaneously: when the target appears, where it is, and what happened, and to also examine the joint use of visual and auditory evidence.64% similarUnverifiedThe CapMem benchmark dataset contains 75 videos with a total duration of 33.7 hours, 1,000 manually annotated multiple-choice questions, covering 16 different scenarios.64% similarUnverifiedOpenMontage 内置了超过 100 种视频制作工具，覆盖剪辑、转码、特效、字幕、音频处理等需求63% similarUnverifiedThe GraftSet-Video dataset constructed by the research team contains 2,397 prompts corresponding to 9,588 format-matched video examples, with an average of approximately 4 examples per prompt.63% similarUnverifiedOmniCapBench contains 786 densely annotated videos.62% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1028082API
curl https://kongchang.com/api/v1/knowledge/claims/1028082MCP
get_claim(id=1028082)